China’s National Supercomputing Center in Shenzhen has unveiled LineShine, a new high-performance computing project that is expected to reach 2 exaflops once fully built out. The system is being promoted as a fully domestic stack, with Chinese-designed processors, storage and networking at its core, as Beijing continues to push for greater self-reliance in advanced computing.

According to reporting by HPCwire and DatacenterDynamics, the first stage of the rollout includes 100 Huawei Kunpeng servers, which provide 12,800 CPU cores. The larger second phase is intended to add many more processors, more capable interconnects and denser storage, with DatacenterDynamics saying the full system is slated to use about 47,000 CPUs across 92 compute cabinets, alongside a large-scale network and storage fabric.

Technical descriptions of the project suggest that LineShine is being built as a distributed ARM-based machine designed to combine strong memory bandwidth with rapid internal communication. That matters because the platform is meant to serve two increasingly important demands at once: conventional scientific modelling and AI training. In that sense, the project reflects a broader shift in supercomputing towards systems that can handle both simulation and machine learning on the same infrastructure.

One of the more notable aspects of LineShine is its CPU-only approach. While many leading supercomputers in the United States and elsewhere depend heavily on GPUs, Shenzhen’s centre is presenting this as an all-CPU alternative built around domestic hardware. Chinese officials have not said when the full installation will be finished or when it will begin operating at its expected peak.

Source Reference Map

Inspired by headline at: [1]

Sources by paragraph:

Source: Noah Wire Services